Sanjay Srinivasa
Machine Learning Engineer
I build and fine-tune ML systems: from diffusion models and RAG pipelines to LLM alignment and production-scale inference. Currently at Corsair Gaming, working on generative AI, multi-agent systems, and intelligent automation. Previously at Optum, building healthcare AI across NLP, computer vision, and large-scale data pipelines.
Passionate about machine learning, AI-driven automation, and building systems that remove toil - always exploring what's next in the space.
Currently
Machine Learning Engineer at Corsair Gaming — Milpitas, CA
Architecting multi-agent systems for AI-powered image generation. Fine-tuning Qwen diffusion models via LoRA and benchmarking DoRA across adapter strategies. Building supply chain planning automations that cut planner cycle time by ~70%. Deploying on-prem LLM inference at scale with vLLM and Docker.
Experience
Corsair Gaming Inc.
Milpitas, California
Machine Learning Engineer
Dec 2025 — Present- Architected a multi-agent system using crewAI for AI-powered image generation, coordinating prompt routing, tool selection, and multi-step reasoning across diffusion and editing models.
- Fine-tuned Qwen diffusion models via LoRA with a two-stage architecture, caching multimodal conditioning to train only the denoiser on precomputed latents; used FP8 quantization, BF16 compute, and gradient checkpointing to fit per-GPU VRAM constraints across dual-GPU on-prem infrastructure.
- Benchmarked DoRA (magnitude/direction decomposition) against LoRA across ranks 4–32 in PyTorch to optimize adapter strategy; validated zero-overhead inference via weight merging.
- Built a supply chain planning automation via DAX reads, lead time mismatch validation, and WebADI-compliant output generation; delivered a dashboard for cycle tracking, cutting planner cycle time by ~70%.
- Designed a demand projection automation sampling historical order velocity for supply forecasts; routed approvals via Graph API and Power Automate with a forecast-vs-actuals dashboard, saving 720 hours/year.
Machine Learning / DS Intern
Jun 2025 — Dec 2025- Built an enterprise RAG pipeline over 200K+ documents using transformer-based embeddings and metadata-driven retrieval; achieved 95% Recall@5 and sub-second latency at scale.
- Deployed a scalable on-prem LLM inference microservice (vLLM, Docker) over an 8M+ vector index; benchmarked throughput and latency under concurrent load to validate production scalability.
Optum (UnitedHealth Group)
India
Associate AI/ML Engineer
Nov 2023 — Aug 2024- Fine-tuned Llama-2-70B on a 4×H100 80GB GPU cluster using DeepSpeed ZeRO-3 and LoRA for clinical decision support and medical reasoning; deployed on Azure AKS with auto-scaling, increasing NPS score by 55%.
- Built LLM alignment pipeline, QLoRA (NF4 4-bit, double quantization) for Supervised Fine-tuning followed by DPO for preference alignment; validated with 34-test suite on quantization fidelity and loss correctness.
- Modeled ensemble methods (XGBoost, Random Forest, Logistic Regression) validated via holdout experiments, improving debt recovery by 70%.
- Engineered a distributed PySpark + SQL pipeline on Databricks to process 100K+ healthcare documents with automated PHI de-identification via SparkNLP, achieving 98.6% precision.
Data Scientist
Mar 2023 — Nov 2023- Designed DBSCAN-based clustering to detect anomalies in patient claims, achieving 98.5% precision.
- Fine-tuned pretrained summarization models on call transcripts, deployed via Azure ML endpoints, reducing handling time by 67% across 10M+ transcripts in production.
Software Engineer - Machine Learning
Jul 2020 — Mar 2023- Fine-tuned YOLOv5 and built an OCR pipeline deployed via ONNX and Triton Inference Server for GPU-accelerated extraction of patient vitals from unstructured PDFs, achieving 93% mAP@0.5.
- Deployed a Dockerized audio-to-text transcription microservice on Azure with Jenkins CI/CD, enabling parallel batch processing and reducing latency by 35%.
Education
University of California, Riverside
Master of Science, Computer Science — Coursework: ML, NLP & DL — GPA: 3.71
Sept 2024 — Dec 2025
Riverside, USA
R.V. College of Engineering (RVCE)
Bachelor of Engineering, Computer Science
Aug 2016 — Jun 2020
Bangalore, India